The narrow passageways and limited visibility in large underground parking lots often lead to vehicle collisions at blind intersections, posing significant dangers. In order analyze approach behavior at these intersections, a driving simulation experiment was designed. The experiment involved constructing a 3D model of the underground parking lot and integrating it with a connected vehicle warning information system. Four experimental scenarios were devised, considering variations in technical features (with or without warning information) and events (with or without vehicle conflicts at blind intersections). Using micro-driving behavior data from 31 participants, key metrics such as speed, acceleration, and braking position were selected to analyze approach behavior from the warning point to the blind intersection. Subsequently, correlation analysis was conducted, followed by the application of the k-means method to cluster driver types and examine the effects of warning information on the approach behavior of different drivers. Finally, the utility of the system was evaluated. The results indicate the following. ①When drivers approached blind intersections without warning information, the process typically involved three stages: speed maintenance, speed increase, and emergency braking. In contrast, when warning information was provided, speed decreased more uniformly and was 34.08% higher than without warning information. ②The warning information system reduced the average speed by 9.94 km/h compared to scenarios without warnings, and advanced the braking position by 4.49 meters, thereby effectively enhancing the safety of drivers passing through blind intersections in parking lots. ③The warning information system had discernible effects on conservative and general drivers, suggesting the need for additional training for radical drivers to help them understand the positive role of the warning system in improving driving efficiency and promoting safe driving practices. ④The warning information system significantly improved overall driver safety, with the greatest impact observed among conservative drivers, followed by ordinary and aggressive drivers. These research findings provide support for the application of connected vehicle warning information systems in parking lots and contribute to the enhancement of parking lot safety.